Cancer Progress and Priorities: Prostate Cancer

Abstract

Prostate cancer is the second most common cancer globally among men, although incidence and mortality rates vary greatly between countries and there are large disparities in prostate cancer incidence and mortality within countries. Despite the high burden of this disease, epidemiologic studies have identified few consensus risk factors for total prostate cancer: age, race, family history, and genetic risk loci. High-grade prostatic intraepithelial neoplasia (HGPIN) is likely a precursor lesion associated with invasive prostate cancer, although the relationship of HGPIN to prostate cancer is complex and remains incompletely understood (1). However, increasing focus on prostate cancer subtypes in epidemiologic research could yield additional insights into prostate cancer etiology and progression that may lead to identification of further consensus risk factors.An estimated 1.3 million cases of prostate cancer were diagnosed in 2018 making prostate cancer the second most common cancer among men globally (2). There is approximately 25-fold variability in incidence rates, with the highest incidence in western and northern Europe, North America, and Australia/New Zealand, intermediate incidence in eastern Europe, South America, southern Africa, and western Asia, and the lowest incidence in southern and eastern Asia, and the rest of Africa (Fig. 1). Some of this heterogeneity can be attributed to variable use of prostate-specific antigen (PSA) testing across countries. Patterns of prostate cancer mortality differ greatly from those for incidence, with the highest mortality rates observed in sub-Saharan Africa, the Caribbean, and South America. Mortality rates are intermediate in Europe, North and Central America, and Australia/New Zealand, and lowest in Asia (Fig. 2). Estimated lifetime risk is estimated to be 6% globally, but ranges from 2% in countries with a low/intermediate sociodemographic index to 14% in countries with a high sociodemographic index (3).Prostate cancer is the most commonly diagnosed cancer among men in the United States (excluding non-melanoma skin cancer) with an annual age-standardized incidence rate of 112.6 per 100,000, as estimated by the Surveillance, Epidemiology, and End Results Program (SEER) between 2011 and 2015 (4). Lifetime risk of prostate cancer is estimated to be 11.2% for U.S. males. Prostate cancer is rarely diagnosed before age 45, although incidence steadily rises from age 45 through age 70, and then declines after age 70 (Fig. 3). The median age of diagnosis is 66 years in the United States. As shown in Fig. 4, incidence rates in the United States approximately doubled from the late 1980s into the early 1990s with the introduction of PSA testing, before stabilizing between 1995 and 2005, and falling since that time.The estimated age-standardized prostate cancer mortality rate is 19.5 per 100,000 in the United States. In part, due to the effects of widespread PSA testing, 5-year survival for prostate cancer is 98% overall, although varies substantially by tumor stage, with 100% 5-year survival for localized (78% of cases) and regional (12%) staged cancers, and 30% 5-year survival for metastatic cancers (5%). Mortality rates steadily grew through the mid-1990s but have declined by 3% to 4% annually since their peak (Fig. 4), likely driven, in part, by both early detection and advances in treatment (5, 6). Five-year survival has increased from 66% in 1975 to its current level.Following the U.S. Preventive Services Task Force (USPSTF) Grade D recommendation against PSA testing in 2012, there was a reduction in PSA testing across age groups and decrease in the incidence of early-stage prostate cancer (7–10). However, rates of advanced and metastatic disease appear to be increasing following this recommendation (10, 11). The USPSTF recommendation for men ages 55 to 69 was reclassified to Grade C (recommendation to selectively offer testing) in 2018, and it is too early to tell what the long-term effects of changes in PSA testing patterns will have on prostate cancer mortality in the United States.There are substantial differences in prostate cancer rates across racial/ethnic groups in the United States that represent important disparities in prostate cancer risk and outcomes (12). At almost every point along the prostate cancer continuum and for most every age group, prostate cancer is more common in African American men than White men. Autopsy studies reveal higher prevalence of latent HGPIN and prostate tumors among African American men relative to White and Asian men (13, 14). The disparity is apparent even in latent prostate cancer, which is detected in autopsies in 50% of men of Asian descent ages 90 to 99, 50% of men of European descent ages 80 to 89, and 50% of men of African descent men ages 60 to 69 (14).There is a similar disparity in the prevalence of cancers detected upon initial screening: The rate of prostate cancer detection in African American men is consistently higher than in White men in the United States (15–19). In addition, the prevalence of screen-detected cancers in Ghanaian men is higher than that reported in any African American population, suggesting the rates of prostate cancer in Africa may equal or exceed those in African Americans (20), and that elevated prostate cancer rates are a phenomenon of men of African ancestry throughout the African diaspora.It is also well known that population-based prostate cancer incidence and mortality vary substantially by race/ethnicity. SEER 18 data from 2011 to 2015 estimated that age-adjusted prostate cancer incidence in African Americans (178.3 per 100,000 men) is substantially greater than in non-Hispanic Whites (105.7 per 100,000 men). This represents 69% greater incidence in African American versus non-Hispanic White men. Prostate cancer mortality statistics show even greater disparities: Mortality in African American men is 39.9 per 100,000 compared with 18.2 per 100,000 in non-Hispanic Whites, a 2.2-fold higher rate. This is the largest African American:White disparity in cancer mortality of any tumor site in U.S. men or women. Despite the magnitude of these prostate cancer racial disparities, they represent an improvement from the disparities that existed only a few years ago. As shown in Fig. 4, the disparity in prostate cancer mortality grew during the period from the mid-1980s to approximately 2005, during which PSA testing became widespread. Since 2010, the absolute disparity in mortality rates between African American and non-Hispanic Whites has declined, though a significant disparity persists. The disparity in prostate cancer mortality is age-dependent, with greatest disparities in prostate cancer mortality occurring in men under age 75.These data suggest that the disparity may have a biological component, as the disparity is evident even before cancer is usually clinically detected. However, the African American:White disparity increases in magnitude in clinically detected disease and mortality, suggesting that factors related to exposure, behavior, or access to care are also important factors in prostate cancer disparities. While it is clear that there is substantial disparity at all phases of the prostate cancer continuum, data suggest that there is racial/ethnic disparity in certain prostate cancer–related parameters but not others. A meta-analysis by Evans and colleagues (2008) reported no disparity in overall survival by race, but did find evidence for a difference in prostate cancer–specific survival and biochemical (PSA) failure that persisted after adjustment for comorbidities, PSA testing, or access to free health care (21). Similarly, it has been suggested that racial differences in prostate cancer mortality diminish or disappear in certain patient subgroups or if treatment is equalized among all patients (22, 23). Studies within the supposedly equal-access Veterans Affairs (VA) healthcare system have found no differences in prostate cancer outcomes or even better outcomes for African American veterans (24–27). However, not all studies have been clearly able to demonstrate that equal access and treatment leads to equal outcome. An analysis of SEER-Medicare data from patients with clinically localized prostate cancer found that mortality rates varied within screening, treatment, and racial groups (28). The data available to date do not completely resolve the question whether racial disparities could be eliminated if treatment were equalized among all cases.Finally, most studies of prostate cancer disparities are based on self-identified race or ethnicity classifications. However, genomically determined ancestry is also finding a place in the study of disparities. The use of genomic ancestry may be able to define groups differently than self-identified race or ethnicity and thus may be a useful adjunct in our understanding of the biological and social determinants of prostate cancer risk, outcomes, and disparities.Prostate cancer is clinically heterogeneous, and while a fraction of tumors is phenotypically aggressive, the majority are indolent. Advanced cancers are variably defined and can refer to higher grade or stage, metastatic, or lethal prostate cancer. Several potential prostate cancer risk factors, such as height, demonstrate stronger associations with the incidence of aggressive disease, while others are solely associated with this phenotype, such as obesity (29, 30).Genomic profiling of prostate tumors in The Cancer Genome Atlas revealed seven molecular subtypes of cancers defined by the presence of ETS fusions or mutations in SPOP, FOXA1, and IDH1 (31). To date, this taxonomy of prostate cancer has had limited application in etiologic epidemiologic research. A notable exception is the prostate cancer subtype defined by presence of the TMPRSS2-ERG fusion, a somatic gene fusion in which the ERG oncogene becomes androgen-regulated (32, 33). Select risk factors, including obesity and height, among others, have been found to be associated with the incidence of TMPRSS2-ERG–positive disease (34–38). There is substantial intratumoral genomic heterogeneity within multifocal prostate tumors, representing an important consideration when evaluating molecular classifications of prostate cancers (39–41).The proportion of prostate cancer that can be explained by known risk factors is one of the lowest of all common cancers (42). The heritability of prostate cancer is among the highest across cancer sites—estimates from twin studies range from 42% to 57% (43, 44). However, migrant studies support a substantial role of lifestyle and environmental factors in prostate cancer etiology (45–47). As summarized in Table 1, the only clear and consistent risk factors identified to date for total prostate cancer are age, race, and family history, with height as a probable risk factor. There is a dearth of modifiable risk factors with convincing evidence for an association with prostate cancer risk (48). Some factors, including obesity, have probable associations with prostate cancer subtypes, such as advanced prostate cancer (often defined as high grade or stage) or lethal prostate cancer. The effects of widespread PSA testing further complicate the identification of risk factors for total or aggressive prostate cancer. PSA-based screening can lead to overdiagnosis of otherwise indolent tumors, and can confer a 3- to 10-year lead time for detection of aggressive tumors, influencing the age and stage at their diagnosis (49). Moreover, undergoing PSA testing is correlated with engaging in healthful behaviors and healthcare utilization, and is a strong predictor of prostate cancer incidence; thus, it is a major potential source of bias in epidemiologic studies of prostate cancer risk (50–52). Further understanding of prostate cancer heterogeneity and consideration of detection biases may yield additional insights into prostate cancer etiology and progression.Body fatness is not associated with total prostate cancer, although the totality of evidence suggests that increased body fatness is associated with a higher risk of advanced prostate cancer. The 2014 World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) Continuing Update Project (CUP) meta-analysis reported an 8% increase in risk of advanced disease per 5-kg/m2 increase in body mass index [BMI; relative risk (RR) = 1.08, 95% confidence interval (CI), 1.04–1.12; ref. 53], in line with findings from other recent meta-analyses (54–56). An umbrella review of adiposity and cancer characterized the evidence for an association between BMI and risk of advanced prostate cancer as weak (57). Alternative measures of adiposity including waist circumference and waist-to-hip ratio are also associated with risk of advanced disease (53). Multiple mechanisms have been proposed for how obesity may contribute to the incidence of advanced prostate cancer, including increased levels of circulating growth factors, induction of chronic inflammation, and reduction of circulating androgen levels (58–60). The association with adiposity is further complicated by potential detection biases, as obese men tend to have lower PSA levels, which may delay diagnosis or make them less likely to undergo biopsy (61, 62).Epidemiologic evidence generally supports an association between greater adult height and higher risk of prostate cancer. The 2014 WCRF/AICR meta-analysis found that each 5-cm increase in adult attained height was associated with a 4% increase in total prostate cancer risk (RR = 1.04; 95% CI, 1.03–1.05; ref. 53) This finding was consistent when examining risk of nonadvanced, advanced, and lethal prostate cancer. Adult height is a product of early life growth rate, which is influenced by levels of hormones and growth factors, including IGF (63–65).A 2016 meta-analysis of 27 studies found an 18% increase in prostate cancer morbidity and mortality comparing higher volume drinkers (65+ g/day) to abstainers (RR = 1.18; 95% CI, 1.10–1.27), with a significant dose–response relationship (66). However, prior meta-analyses found no significant association and the WCRF/AICR found no conclusive evidence for an association (53, 67, 68). In addition, several prospective studies have reported inverse associations between alcohol consumption and the risk of advanced or lethal prostate cancer (69–71).While smoking has not been linked to prostate cancer incidence, there is suggestive evidence that smoking is associated with higher prostate cancer mortality, with an estimated 15% to 25% higher rate of prostate cancer–related death among current smokers (72–74). Adding to the complexity of this association is that smokers are less likely to undergo PSA testing, leading to later diagnosis and higher incidence of aggressive cancers (50, 75). Smoking may also lead to worse prostate cancer outcomes postdiagnosis, potentially predicting worse response to radical prostatectomy and radiotherapy (76).A recent meta-analysis of 72 studies (77) found no association between overall physical activity and total prostate cancer incidence, although prior meta-analyses found modest inverse associations (78). An umbrella review of leisure time physical activity and cancer found a modest increase in total prostate cancer risk (79). The WCRF/AICR determined that evidence for an association was inconclusive (53). The strongest evidence in support of an inverse association is from several large prospective cohort studies that found higher levels of physical activity, particularly vigorous physical activity, were associated with approximately 30% lower risk of advanced prostate cancer (80, 81). The heterogeneity in these findings may be explained by variable classifications of physical activity (recreational vs. occupational, type and intensity of physical activity, etc.) as well as by tumor heterogeneity in prostate cancer (82). Physical activity is hypothesized to act on prostate cancer risk and progression through effects on hormone levels, inflammation, and growth factors (82).Dairy consumption is suggestively associated with increased prostate cancer risk. In concordance with prior meta-analyses, the WCRF/AICR found a 7% increase in risk per each 400 g intake per day (RR = 1.07; 95% CI, 1.02–1.12; refs. 53, 83, 84). Findings were nonsignificant when evaluating advanced and fatal disease. Correlated with dairy intake, a 400 mg/day increase in dietary calcium intake was associated with a 5% increase in total prostate cancer risk in the WCRF/AICR report (RR = 1.05; 95% CI, 1.02–1.09)–again, supported by prior meta-analyses (53, 84, 85). However, there was no association when evaluating risk of advanced prostate cancer (53). Despite the suggestive association with calcium, there is only limited evidence for an association between dietary vitamin D or circulating levels of vitamin D (25(OH)D) and prostate cancer risk (53, 85–87).There is suggestive evidence that diets low in selenium intake are associated with increased prostate cancer risk (53). Meta-analyses examining total prostate cancer risk have found approximately 25% lower risk comparing highest with lowest quantiles of serum selenium concentrations (88, 89). These findings were suggestively stronger when evaluating advanced prostate cancer. However, selenium supplementation in the SELECT randomized trial did not reduce prostate cancer risk (90). A majority of participants in SELECT had relatively high baseline concentrations of selenium, possibly masking any true prostate cancer–protective effect that selenium supplementation may have on a selenium-deficient population. Subsequent stratified analyses showed that selenium supplementation was associated with increased risk of high-grade prostate cancer among men with high baseline selenium concentrations (91).The carotenoid lycopene, a potent antioxidant commonly found in in tomatoes and tomato-based products, may play a protective role in prostate cancer. High dietary consumption of lycopene is associated with approximately 10% lower prostate cancer risk in recent meta-analyses, although the WCRF/AICR deemed evidence inconclusive (53, 92, 93). This association was consistent when evaluating lycopene levels in circulation (92). There were suggestive, but nonstatistically significant, inverse associations between dietary and circulating lycopene levels and aggressive, advanced, and lethal prostate cancer (92).There is no evidence for an association between beta-carotene consumption or supplementation and total or advanced prostate cancer risk (53).Consumption of 6 cups of coffee per day (versus no consumption) was associated with 60% lower risk of lethal prostate cancer in the Health Professionals Follow-Up Study (94). A subsequent meta-analysis supported this finding, concluding that coffee consumption is associated with approximately 10% decrease (per each 3 cups/day) in the incidence of high-grade and fatal prostate cancer, while there is no evidence for an association with total disease (95).There is growing evidence that use of statins, a class of lipid-lowering medications, may be associated with lower prostate cancer risk, particularly for advanced disease. A 2012 meta-analysis of 27 observational studies reported relative risks of 0.93 (95% CI, 0.87–0.99) and 0.80 (95% CI, 0.70–0.90) for total and advanced prostate cancer, respectively (96). However, a subsequent meta-analysis did not observe significant associations within any particular type of statin (97). Statin use postdiagnosis is also potentially associated with better survival (98, 99).A 2015 meta-analysis of 56 studies found that men with type II diabetes experienced 12% lower risk of total prostate cancer (RR = 0.88; 95% CI, 0.82–0.93), with mixed findings reported in subsequent large studies (100–102). The evidence remains inconclusive for this association, however, as there is no clear association between glycemic biomarkers and prostate cancer risk (103–106), and metformin use is not associated with prostate cancer risk (107, 108). The evidence also does not support an association between diabetes and risk of advanced or aggressive prostate cancer (102, 109, 110). Preexisting diabetes was associated with 29% higher prostate cancer mortality in a recent meta-analysis (111).The body of evidence is inconclusive for potential associations between vasectomy (112), ejaculation frequency (113), exposure to Agent Orange (114), sexually transmitted infections including Trichomonas vaginalis (115), or prostatitis (116) and prostate cancer risk.In contrast to the limited or inconsistent evidence for epidemiologic risk factors, there is strong evidence that prostate cancer etiology, aggressiveness, and progression are influenced by genetics. Family history of prostate cancer in a father or brother is associated with approximately 2-fold higher risk of prostate cancer (117–119). Prostate cancer has the highest heritability of any common cancer. However, few hereditary prostate cancer syndromes have been reported, and the genes that may explain these have not been confirmed. Family-based linkage studies of hereditary prostate cancer have identified high penetrance genes including HPC1 (1q24–25; refs. 120–122), PCAP (1q42–43; refs. 122–124), HPCX (Xq27–28; ref. 125), CAPB (1q36; refs. 122, 124), HPC20 (20q13; ref. 126), HOXB13 (127, 128), and others However, these have not been into the gene or mutations have not been or the associations have not been to application in prostate cancer risk In mutations in are in risk of a may also be in prostate cancer screening as well as in both early-stage and disease In addition, HOXB13 and gene mutations identified in are clearly associated with prostate cancer and represent for genetic testing and risk the of in patients with mutations and in patients with mutations and a strong for genetic testing in certain men with prostate cancer large of low to penetrance have been associated with prostate cancer in association At common associated with prostate cancer have been reported of relative risk for prostate cancer. of genes that have been identified large gene or the androgen ref. genes that prostate refs. genes refs. and ref. or genes Despite the high risk and prostate cancer outcomes in African American men, most associations reported in European or Asian descent have not been in African descent and few have been identified in African American or African Multiple genomic associations at have been as prostate cancer in including African Americans no gene has been to be for this cancer risk, of the gene or by has been reported risk have been that these low penetrance as well as other the additional of these to risk is particularly for predicting aggressive prostate cancer the of these has not to clearly or that have the in is in for risk and research will be to the application of this and its in early detection and for of prostate cancer risk have been including in the Prostate Cancer and European Study of for Prostate Cancer although are These are to the presence of prostate cancer among men for biopsy and have from to the dearth of strong prostate cancer risk factors, these age, race, family history, PSA levels and of and prior of prostate cancer incidence rates suggest a to 2% annual decrease through in the United States However, the of risk in the United with of prostate cancer are to with the greatest growth among African American men of metastatic disease is to increase by annually with greater increases for men under age 70 Mortality rates are to decrease by approximately 3% annually for prostate cancer have on and lifestyle although large prostate cancer have been The SELECT trial found that vitamin supplementation increased prostate cancer risk by among U.S. but selenium did not risk randomized prostate cancer through treatment with In the initial report of the treatment total prostate cancer incidence by but there was relative increase in the incidence of high-grade disease in the The trial similar with treatment total prostate cancer incidence by but associated with a suggestive increase in the incidence of high-grade disease In of these the a to The effect of on high-grade prostate cancer incidence in of the (RR = 95% CI, ref. In addition, there was no increase in the risk of prostate cancer mortality, worse survival following a diagnosis of prostate cancer in the suggesting that the observed higher incidence of high-grade disease was likely an potentially from detection biases from the effects on prostate volume PSA testing has been commonly for the early detection of prostate cancer since the late although its widespread use is with a of an American Cancer review found that the of the PSA was for high-grade while the was among men at prostate cancer risk widespread prostate cancer screening PSA can lead to overdiagnosis and while to advanced the United States Preventive Services Task Force (USPSTF) testing between men ages 55 to 69 and their while PSA testing is not for men ages 70 or The USPSTF did not but suggested that screening at ages may be for African American men and men with a family history of prostate cancer. large randomized have been to the of PSA testing on the reduction of prostate cancer the and Cancer the and the of PSA for Prostate Cancer ref. in these show that PSA testing leads to a higher incidence of prostate cancer, but the had variable findings prostate cancer found a significant reduction in mortality, but no mortality was observed in in early from However, was limited by of the there was low in the and there was substantial heterogeneity in findings across study in show that upon adjustment for differences in trial and from and were both with a 25% to 30% reduction in prostate cancer mortality Prostate cancer screening may be through of baseline screening of other serum such as free the and the Prostate Health or use of such as or The increasing use of for of prostate cancers may reduce the potential for associated with PSA-based screening A major in the early detection of prostate cancer is to that can be in groups of men and to based on a screening to and such as may in the of advanced, and metastatic disease, and the of while could diagnosis of prostate cancer in prostate cancer research will be to further our understanding of tumor research focus on the of indolent and aggressive tumors, and further the etiology of the of tumor molecular subtypes may lead to risk as well as evidence potential prostate cancer risk In addition, consideration of detection biases in the study and of epidemiologic studies will be for evidence potential of prostate cancer incidence and further study of the and of prostate cancer disparities will be to the higher burden of prostate cancer among African American potential of were and of of data and and of data of the or support or was supported in by and of of this were in by the of This be in with 18 solely to this